Zinc transporter 1 (ZNT1) expression on the cell surface is elaborately controlled by cellular zinc levels

Zinc transporter 1 (ZNT1) is the only zinc transporter predominantly located on the plasma membrane, where it plays a pivotal role exporting cytosolic zinc to the extracellular space. Numerous studies have focused on the physiological and pathological functions of ZNT1. However, its biochemical features remain poorly understood. Here, we investigated the regulation of ZNT1 expression in human and vertebrate cells, and found that ZNT1 expression is posttranslationally regulated by cellular zinc status. We observed that under zinc-sufficient conditions, ZNT1 accumulates on the plasma membrane, consistent with its zinc efflux function. In contrast, under zinc-deficient conditions, ZNT1 molecules on the plasma membrane were endocytosed and degraded through both the proteasomal and lysosomal pathways. Zinc-responsive ZNT1 expression corresponded with that of metallothionein, supporting the idea that ZNT1 and metallothionein cooperatively regulate cellular zinc homeostasis. ZNT1 is N-glycosylated on Asn299 in the extracellular loop between transmembrane domains V and VI, and this appears to be involved in the regulation of ZNT1 stability, as nonglycosylated ZNT1 is more stable. However, this posttranslational modification had no effect on ZNT1's ability to confer cellular resistance against high zinc levels or its subcellular localization. Our results provide molecular insights into ZNT1-mediated regulation of cellular zinc homeostasis, and indicate that the control of cellular and systemic zinc homeostasis via dynamic regulation of ZNT1 expression is more sophisticated than previously thought.

In vertebrates, cellular and systemic zinc homeostasis are maintained through its mobilization across the membranes by two zinc transporter family proteins, ZIP 3 and ZNT (1-3), which are located on the membranes of both the cell surface and intracellular compartments. Most ZIP proteins are on the plasma membrane and mobilize zinc from the extracellular space to the cytosol as zinc importers, and therefore, cells have a number of zinc entry routes (1)(2)(3). In contrast, most ZNT proteins are in intracellular compartments, and ZNT1 is the only ZNT protein that predominantly functions on the plasma membrane as a zinc efflux transporter (4 -6), although other ZNT proteins can localize to the cell surface (7)(8)(9). Therefore, ZNT1 is a crucial regulator of cellular zinc homeostasis.
Discovered in 1995, ZNT1 was the first mammalian zinc transporter protein identified (4), and it is now known to play important roles in physiological and pathophysiological processes (10). ZNT1 protects cells from zinc toxicity by exporting cytosolic zinc into the extracellular space (11)(12)(13), which may be important for protecting neurons after transient forebrain ischemia (14,15). Its functions in controlling cytosolic zinc levels regulate the activation of RAF-1, which is a signal transducer in the RAS-ERK pathway (16), which may protect cells from ischemia reperfusion (17). The functions of ZNT1 that control signaling are conserved among ZNT1 orthologs, because a Caenorhabditis elegans ortholog controls RAS-ERK signaling (18). ZNT1 is essential for embryonic development because it transports maternal zinc into the embryonic environment, and homozygous znt1 knockout mice exhibit early embryonic death (19). In enterocytes, ZNT1 is located on the basolateral membrane, so probably facilitates zinc absorption by exporting it into portal blood (6,10,20). In addition to its functions as a zinc exporter, ZNT1 also plays regulatory roles, most of which are related to protein-protein interactions with ZNT1 in intracellular compartments. In the endoplasmic reticulum (ER), ZNT1 interacts with the ␤-subunit of the L-type calcium channel, which leads to a reduction in the cell surface expression of its pore-forming ␣ 1 -subunit (21). Moreover, its interaction with EVER proteins may be involved in the pathogenesis of Epidermodysplasia verruciformis, which is a rare autosomal recessive skin disease (OMIM 226400) (22). Taken together, these observations clearly highlight ZNT1's importance in cellular and systemic homeostatic control. However, the molecular feature of ZNT1 is relatively unknown.
A well-known molecular feature of ZNT1 is that it is transcriptionally up-regulated in response to high zinc levels, and this is mediated through the binding of metal-response element-binding transcription factor-1 (MTF-1) to the metal-response elements in its promoter region (23)(24)(25). This transcriptional regulation in response to high zinc levels is similar to that of metallothionein (MT), which is also a target gene of MTF-1. However, other features of ZNT1 expression regulation still remain poorly understood. In this study, we investigated the molecular features and expression regulation of ZNT1 in human and vertebrate cells using an anti-ZNT1 mAb that we generated in previous studies (26).

ZNT1 is N-glycosylated on the Asn 299 residue, which neither affects the ability to confer resistance against high zinc levels nor the subcellular localization of ZNT1
We first confirmed whether our anti-ZNT1 mAb, which was generated against the C-terminal cytosolic portion of the protein, specifically detects ZNT1 using an N-terminally FLAGtagged ZNT1 (FLAG-ZNT1) protein, because it detected ZNT1 at a greater molecular size than that calculated based on its cDNA in our previous study (26 -28). The anti-ZNT1 mAb detected FLAG-ZNT1 stably expressed in DT40 cells deficient in znt1, mt, and znt4 (znt1 Ϫ/Ϫ mt Ϫ/Ϫ znt4 Ϫ/Ϫ ) (27,29) as two bands (mainly ϳ75 and ϳ63 kDa) in immunoblotting (Fig. 1A,  left), which was also detected in the same manner by an anti-FLAG antibody (Fig. 1A, right), indicating that both bands were specific to ZNT1. We conducted two experiments to investigate whether the presence of two bands may have been caused by variations in glycosylation. First, we treated a cell lysate prepared from znt1 Ϫ/Ϫ mt Ϫ/Ϫ znt4 Ϫ/Ϫ cells overexpressing FLAG-ZNT1 with PNGase F, which hydrolyzes all types of N-linked oligosaccharide structures. The PNGase F treatment shifted the ϳ75 kDa band of ZNT1 to ϳ63 kDa, indicating that the higher band is the N-glycosylated form of ZNT1 (Fig. 1B, lane  4). The ϳ63 kDa band was still higher than 55.3 kDa, the expected molecular mass of ZNT1, which could be attributed to the properties of the cytosolic histidine-rich loop. Second, because ZNT1 has a consensus motif for N-glycosylation (Asn 299 -Ser 300 -Thr 301 ) in extracellular loop 3 (EL3) between transmembrane domains V and VI, we constructed a nonglycosylated ZNT1 mutant in which the 299th Asn residue (Asn 299 ) was substituted with an Ala residue (ZNT1 N299A ). Immunoblotting of ZNT1 N299A only showed the ϳ63 kDa band (Fig. 1B, lane 5), clearly indicating that ZNT1 is N-glycosylated at Asn 299 .
Then, we investigated the effect of N-glycosylation at Asn 299 on ZNT1 functions, specifically its subcellular localization and ability to tolerate high zinc concentrations. The cell surface localization of the ZNT1 N299A mutant was confirmed by immunofluorescence staining (Fig. 1C, upper panels) and a cell surface biotinylation assay (Fig. 1C, lower panels), indicating that N-glycosylation at Asn 299 is unimportant for the subcellular localization of ZNT1. Because our previous results demonstrated that znt1 Ϫ/Ϫ mt Ϫ/Ϫ znt4 Ϫ/Ϫ cells can be used to evaluate ZNT1 zinc transport ability (27,29), we examined the ability of the ZNT1 N299A mutant in these cells. The ZNT1 N299A mutant conferred resistance to znt1 Ϫ/Ϫ mt Ϫ/Ϫ znt4 Ϫ/Ϫ cells against high zinc concentrations, comparable with that of wildtype (WT) ZNT1 (Fig. 1D). These results indicate that N-glycosylation is unimportant in terms of ZNT1's ability to traffic to the cell surface or to confer cellular resistance against high zinc levels.

Characterization of endogenous ZNT1 in cultured human cells
Previous studies have revealed that ZNT1 plays crucial roles in reducing zinc toxicity when overexpressed (11)(12)(13), but its molecular characterization remains to be elucidated. We characterized ZNT1 using cultured human cells, in which the ZNT1 gene was disrupted by CRISPR/Cas9-mediated genome editing. The loss of ZNT1 in chronic myelogenous leukemia haploid HAP1 or pancreatic cancer PANC1 cells was confirmed by immunoblotting using our anti-ZNT1 mAb (Fig. 2, A and B). The results of the genome editing were confirmed by genomic DNA sequencing (Fig. S1). In ZNT1-deficient cells, MT protein expression was significantly increased compared with that in WT cells (Fig. 2, A and B), suggesting that the loss of ZNT1 increases the cytosolic zinc contents. Treating the total cellular lysates with PNGase F shifted the ϳ75 kDa band of endogenous ZNT1 to the ϳ63 kDa band (Fig. 2, C and D), similar to the above experiment, confirming that endogenous ZNT1 is also N-glycosylated. The re-expression of ZNT1 in ZNT1-deficient PANC1 cells decreased MT induction to the basal level found in WT PANC1 cells but did neither by the expression of ZNT1 H43N and ZNT1 H43A mutants (Fig. 2E), both of which had lost their zinc transport ability, as described previously (26,27). Consistent with these results, ZNT1 re-expression conferred resistance against high zinc concentrations but both the mutants did not (Fig. 2F). These results confirm that ZNT1 plays critical roles in reducing zinc toxicity by effluxing excess zinc.
ZNT1 has been shown to be located on the basolateral membrane (6,10,20), which was confirmed in this study. The cell surface biotinylation assay and Z-stack analysis of the immunofluorescence microscopy results revealed that endogenous ZNT1 was located on the basolateral membrane in polarized Caco2 cells (Fig. 2, G and H) and was N-glycosylated, indicating that N-glycosylation is a general modification of the ZNT1 protein (Fig. 2I).

Zinc-induced ZNT1 accumulation on the cell surface
ZNT1 mRNA expression increases in response to high zinc levels (23)(24)(25)30), which was confirmed in hepatoma HepG2 cells by zinc supplementation (50 M ZnSO 4 ) (Fig. 3A). Consistent with the increase in its mRNA expression, the expression of the ZNT1 protein also increased (Fig. 3B), but the time course of these increases during ZnSO 4 treatment differed. The ZNT1 expression at the protein level increased gradually up to 12 h, whereas its expression at the mRNA level increased for up to 3 h and decreased in 6 h. Therefore, we examined the expression of the ZNT1 protein at the post-translational level in more detail.
ZNT1 protein expression was increased in HepG2 cells by low levels of zinc supplementation (20 M ZnSO 4 ; Fig. 3C)

Zinc-responsive ZNT1 expression
and in response to increases in zinc concentration (Fig. 3D). MT expression increased with that of ZNT1, but was higher than that of ZNT1 (Fig. 3, C and D). Similar responses of both proteins were confirmed in PANC1 cells (Fig. 3, E and F). Therefore, ZNT1 protein expression as well as mRNA expression (23,25,30), increases in response to an increase in zinc concentration. The cell surface biotinylation assay of HepG2 cells revealed that ZNT1 accumulation on the plasma membrane increased in response to increased zinc (Fig. 3G), consistent with its zinc-efflux function.
We then examined the possibility that ZNT1 accumulation on the plasma membrane was directly facilitated by increased zinc, not just by increases in translation following its transcriptional up-regulation in response to increases in zinc. We used mtf-1 knockout (mtf-1 KO) MEF cells, because we assumed that the effects of zinc on ZNT1 cell surface accumulation are directly evaluated in mtf-1 KO MEF cells, which have lost cellular responses to high zinc toxicity through up-regulating the transcription of a set of zinc-responsive genes, such as MT and ZNT1 (31,32). Culturing in high zinc concentrations (80 M Note that nonglycosylated ZNT1 proteins of the same size were detected in lanes 4 and 5. In A and B, calnexin (Cnx) and tubulin were used as loading controls, respectively. C, ZNT1 N299A mutant located on the plasma membrane, as in wildtype (WT) ZNT1. Immunofluorescence staining confirmed that the ZNT1 N299A mutant was located on the cell surface, as was WT ZNT1 (upper two panels). A cell surface biotinylation assay confirmed this (other panels). Input refers to aliquots of the biotinylated proteins before avidin capture (i.e. total cell lysate), and biotinylation refers to avidin-captured proteins. Tubulin and IgM were used as loading controls for input and biotinylation, respectively. D, ZNT1 N299A mutant retained the ability to confer resistance against high zinc concentrations. WT DT40, ⌬1M4, ⌬1M4 cells overexpressing WT ZNT1, and ⌬1M4 cells overexpressing ZNT1 N299A were grown in the presence of the indicated concentrations of ZnSO 4 for 2 days, and the number of live cells was evaluated by an AlamarBlueா assay. Relative values are plotted as a percentage of live cells without ZnSO 4 for each group of cells. Each experiment was performed at least three times, and representative results are displayed. ZnSO 4 or higher concentrations) significantly decreased the viability of mtf-1 KO MEF cells, which was reversed by constitutive ZNT1 expression (Fig. 3H). We then examined whether the accumulation of ZNT1 to the plasma membrane was facilitated by treating the cells with 20 M ZnSO 4 after culture in zinc-deficient medium for 48 h to remove accumulated ZNT1 from the plasma membrane. Although the total ZNT1 protein expression level did not change in response to ZnSO 4 treatment (Fig. 3I, input), the cell surface ZNT1 protein expression level increased after 3 and 6 h (Fig. 3I, biotinylation), clearly indicating the facilitated accumulation of ZNT1 on the cell surface in response to increases in zinc.

ZNT1 accumulated on the cell surface is endocytosed and degraded in the proteasomal or lysosomal pathways under zinc deficiency
While performing the experiments described above, we found that cell surface ZNT1 levels decreased under zinc deficiency. Specifically, ZNT1 expression decreased for up to 48 h of zinc deficiency (Fig. 4A), and the decrease was restored by zinc supplementation in HepG2 cells (Fig. 4B). Similar changes in ZNT1 expression were observed in PANC1 cells (Fig. 4, C and D). The decreases in ZNT1 expression were in concert with those of the MT protein, although MT responses were greater (Fig. 4, A-D). We then explored whether cell surface ZNT1 expression was directly affected by zinc deficiency. ZNT1 expression on the cell surface decreased according to both a cell surface biotinylation assay (Fig. 4E) and immunofluorescence microscopy (Fig. 4F) in zinc-deficient culture for up to 48 h. ZNT1 mRNA expression was almost constant (Fig. 4H), whereas ZNT1 protein expression significantly decreased during zinc deficiency (Fig. 4G).
We then conducted a time course experiment in which ZNT1 degradation was monitored after protein synthesis was blocked by cycloheximide (CHX). ZNT1 degradation was only detected by treating HepG2 cells with CHX ( Fig. 4I, lanes 2-4). The degradation rate of the ZNT1 protein was increased under zinc deficiency, which was caused by treatment with the membrane-permeable zinc chelator TPEN (Fig. 4I, lanes 5-7), which was suppressed by 20 M ZnSO 4 supplementation (Fig. 4I, lanes 8 -10), indicating that physiological levels of zinc inhibit ZNT1 protein degradation. We then examined the effects of lysosome and proteasome inhibitors on ZNT1 degradation. Treatment with TPEN reduced ZNT1 protein expression levels, which were restored by treatment with the proteasomal inhibitor MG132 or lysosomal inhibitor bafilomycin A1 (Fig. 4J, upper panel), indicating that ZNT1 is degraded in both cellular deg-radation pathways during zinc deficiency. In this experiment, MG132 and BafA1 restored ZNT1 expression, but not the amount of ZNT1 on the cell surface (Fig. 4J, lower panel), suggesting that the ZNT1 that accumulates on the cell surface is endocytosed and then degraded. This is indeed the case, because blocking endocytosis by culturing the cells in a medium containing 350 mM sucrose suppressed ZNT1 protein degradation in both total cellular lysates and cell surface proteins (Fig.  4K). These results suggest that the ZNT1 protein expression level is controlled in a sophisticated manner to minimize zinc efflux from cells under zinc defeciency.

N-Glycosylation at Asn 299 is involved in ZNT1 degradation
The above results indicate that ZNT1 protein expression is controlled in various ways. Because one of the unique features of ZNT1 is its N-glycosylation (see Fig. 1), we next examined whether or not it contributes to the regulation of ZNT1 protein expression by zinc status. We used MDCK FLp-In T-Rex (hereafter MDCK) cells stably expressing WT ZNT1 or the ZNT1 N299A mutant to keep their protein expression levels comparable, because transcription was driven by a Tet-regulatable promoter from the same locus in cells harboring the FLp-In T-Rex system (Fig. 5A). First, to monitor the stability of WT ZNT1 and ZNT1 N299A mutant proteins, the cells were cultured with doxycycline (Dox) for 24 h to induce ZNT1 expression, and then cultured for up to 6 h in the presence of CHX to block protein synthesis after Dox was washed out. No significant differences were observed in ZNT1 N299A mutant expression during the time course, whereas marked decreases were found in WT ZNT1 protein expression (Fig. 5B, input). The protein degradation of WT ZNT1 increased in a zinc-deficient culture (Fig. 5C, input), and the cell surface biotinylation assay revealed that ZNT1 on the cell surface decreased in both conditions (Fig. 5, B and C, biotinylation), and significantly decreased under zinc deficiency. These results indicate that N-glycosylation at Asn 299 is involved in ZNT1 degradation and its sensitivity to zinc deficiency.

Discussion
ZNT1 was the first identified mammalian zinc transporter (4). In contrast to its contribution to physiology and pathogenesis (10,19,22), its molecular expression regulation by zinc status has been poorly elucidated, except for its zinc-dependent transcription (23)(24)(25)30). In this study, the clear detection of endogenous ZNT1 by our mAb enabled us to molecularly characterize it, and our findings are summarized as follows. First, ZNT1 protein accumulation on the cell surface increases in Genome editing was confirmed by genomic DNA sequencing (Fig. S1). In A and B, note that MT expression was significantly increased in ZNT1-deficient cells compared with WT cells. C and D, endogenous ZNT1 is N-linked glycosylated in HAP1 (C) and PANC1 (D) cells. Total cellular lysate prepared from cells were treated with or without PNGase F and then subjected to immunoblotting. E, re-expression of WT ZNT1, but not ZNT1 H43N or ZNT1 H43A mutants, restored the decreased MT expression in ZNT1-deficient PANC1 cells. F, reduced resistance to high zinc in ZNT1-deficient PANC1 cells was restored by WT ZNT1, but not ZNT1 H43N or ZNT1 H43A mutants. An AlamarBlueா assay was performed as shown in Fig. 1D. Relative values are plotted as a percentage of live cells without ZnSO 4 for each group of cells. G and H, endogenous ZNT1 was on the basolateral membrane in polarized CaCo2 cells. CaCo2 cells that were grown on Transwellா plates until polarized were subjected to cell surface biotinylation assays (G) or immunofluorescence staining with Z-stack analysis (H). I, ZNT1 on the basolateral membrane in polarized CaCo2 cells was N-glycosylated. Polarized CaCo2 cells were subjected to cell surface biotinylation assays, and biotinylated proteins were treated with or without PNGase F for 24 h before avidin capture. In G and I, input refers to aliquots of the biotinylated proteins before avidin capture, and biotinylation refers to avidin-captured proteins. Na ϩ /K ϩ -ATPase was used as a loading control for input and biotinylation. Each experiment was performed at least three times, and representative results are displayed.

Zinc-responsive ZNT1 expression
response to increases in zinc, which is enhanced by its facilitated accumulation, in addition to transcriptional up-regulation. Second, the ZNT1 accumulated on the cell surface decreases because of degradation in cellular degradation pathways following endocytosis from the cell surface under zinc deficiency. Third, ZNT1 is N-glycosylated at Asn 299 in the EL3 between transmembrane domains V and VI, which is important for its stabilization regulation, despite not affecting ZNT1's ability to traffic to the cell surface nor to confer cellular resistance against high zinc levels. This molecular evidence provides insights into ZNT1 protein functions, considering its multifarious physiological and pathological roles.
ZNT1 mRNA levels increase in response to excess zinc, which is mediated by the binding of MTF-1 to metal-response In A, total RNA prepared from HepG2 cells was subjected to RT-qPCR. Changes in mRNA expression relative to that in a normal medium are shown after normalization with Cyclophilin A mRNA expression. Each value is the mean Ϯ S.D. of three independent experiments (*, p Ͻ 0.05; n.s., not significant). In B, changes in ZNT1 protein expression were evaluated by immunoblotting. C-F, zinc-induced expression of ZNT1 protein in HepG2 (C and D) and PNAC1 (E and F) cells. In C and E, both cells were cultured in medium supplemented with 20 M ZnSO 4 for the indicated period, and in D and F, cells were cultured in a medium supplemented with the indicated concentration of ZnSO 4 for 48 h. In C-F, tubulin is shown as a loading control. G, zinc-induced ZNT1 accumulation on the cell surface. HepG2 cells were cultured in the same manner as shown in C, and cell surface biotinylation assays were performed. Tubulin and transferrin receptor (TFR) were used as loading controls for input and biotinylation, respectively. H, WT ZNT1 expression reversed the zinc-sensitive phenotypes of mtf-1 KO MEF cells, which failed to grow in the presence of 80 M ZnSO 4 or higher concentrations. I, ZNT1 protein displayed facilitated accumulation on the plasma membrane in response to increases in zinc. mtf-1 KO MEF cells were cultured in zinc-deficient medium for 48 h and then cultured for the indicated time after 20 M ZnSO 4 was added. The cell surface localization of the ZNT1 protein was evaluated by a cell surface biotinylation assay. Tubulin and TFR were used as loading controls for input and biotinylation, respectively. In B-G and I, the band intensities of ZNT1 were quantified by densitometric analysis after normalization to the levels of tubulin or TFR, and the ratio relative to the basal condition, which was set to 1.0, is shown below each lane. Each experiment was performed at least three times, and representative results are displayed.

Zinc-responsive ZNT1 expression
elements in ZNT1's promoter (23)(24)(25). Because MT has a similar response, ZNT1 and MT may cooperatively manage cellular responses against zinc toxicity (13). To perform this func-tion efficiently, ZNT1 has been speculated to accumulate on the cell surface (33); however, this idea has not been proven experimentally. We show that ZNT1 protein accumulation on the  was more stable than WT ZNT1 on the cell surface. MDCK cells expressing WT ZNT1 or ZNT1 N299A mutant were treated with 20 g/ml of CHX for 2, 4, or 6 h after treatment with 1.0 g/ml of Dox for 24 h, and cell surface biotinylation assays were performed as described in the legend to Fig. 1C. C, zinc deficiency facilitated the degradation of WT ZNT1 on the cell surface. MDCK cells were cultured as in B, except for adding zinc chelator TPEN, and subjected to the same experiments. In B and C, tubulin and Na ϩ /K ϩ -ATPase were used as loading controls for input and biotinylation, respectively. Each experiment was performed at least three times, and representative results are displayed.

Zinc-responsive ZNT1 expression
cell surface is not only increased by its expression, but also by its facilitated accumulation to the plasma membrane in response to increases in zinc. This novel response is important, because it enables a rapid response to high zinc levels without requiring transcription. The cell surface accumulation of ZNT1 is reminiscent of the copper efflux transporter ATP7A, which accumulates on the plasma membrane from the trans-Golgi network to export copper under high copper conditions (34,35). Phosphorylation is important in the trafficking of ATP7A (36,37). Because ZNT1 has a potential phosphorylation site at Ser 506 in the cytosolic C-terminal (38), similar phosphorylation mechanisms may regulate ZNT1 accumulation. In addition, the facilitated cell surface accumulation of ZNT1 is reminiscent of the cell surface expression of ZIP4, although it responds to zinc in the opposite manner. ZIP4 accumulates on the cell surface through reduced internalization from the plasma membrane during zinc deficiency and is rapidly degraded via endocytosis in response to excess zinc (39,40), but the molecular mechanism behind this has not yet been elucidated. Further investigation is required to elucidate the molecular regulation events underlying the facilitated accumulation of ZIP4 and ZNT1 on the cell surface.
Another significant result of this study is that we found that ZNT1 was degraded via both proteasomal and lysosomal pathways under zinc deficiency. Zinc deficiency triggers endocytosis of cell surface ZNT1, leading to lysosomal degradation, and also potentially the degradation of nascent and intracellular localized ZNT1, through proteasomal degradation via the endoplasmic reticulum-associated protein degradation (ERAD) pathway. This novel response of ZNT1 would be physiologically important for minimizing the zinc efflux from cells during zinc deficiency because it was observed in cells that were made zincdeficient in a physiologically relevant manner using Chelex 100treated fetal calf serum (FCS). Importantly, the degradation of ZNT1 was in concert with that of MT. Considering the concerted degradation of ZNT1 and MT under zinc deficiency, their expression regulation should ensure the use of cytoplasmic zinc in cells during such conditions, and thus be critical for homeostasis maintenance.
As mentioned in the Introduction, ZNT1 is a moonlighting protein. It plays pivotal roles as a zinc exporter on the cell surface and as a binding partner for several proteins in the ER, and operates as a scaffold protein. Our results highlight the zinc-responsive accumulation of ZNT1, which would improve our understanding of the moonlighting functions of ZNT1, e.g. ZNT1 accumulation on the cell surface in response to cellular zinc increases may reduce its scaffold functions. Zinc-responsive expression regulation of ZNT1 may be a novel mechanism of zinc signaling control, considering the important functions of its binding partner proteins, such as RAF-1 and calcium channels. This attractive hypothesis needs further investigation.
For the first time, we present evidence that ZNT1 is N-glycosylated at Asn 299 , and this is involved in stability control but not in zinc efflux functions, as well as its cell surface localization. In a previous study, ZNT2 was shown to be N-glycosylated (41), but its biological significance has not yet been clarified. Therefore, our finding is novel and interesting, in that one can consider the role of extracellular glycosylation in cellular zinc homeostasis. The N-glycosylation site of ZNT1 is conserved among its vertebrate orthologs, but not in the nematode or fruit fly (Table 1). Therefore, fine-tuned ZNT1 expression regulation mediated by N-glycosylation may have only been acquired in vertebrates. The N-glycosylation of metal transporters may be important as a regulatory mechanism for maintaining metal homeostasis, e.g. the glycosylation of ZIP14 is crucial for its degradation control and its sensitivity to iron, so its deglycosylation results in its stabilization (42), as in the case of ZNT1. Similar mechanisms may operate in the degradation control of these two metal transporters. Further investigation is required to clarify this point.
We found that ZNT1 expression on the cell surface is increased by treatment with 20 M ZnSO 4 , and its degradation can be blocked by normal culture medium. These zinc concentrations are physiologically relevant. Moreover, cell surface-accumulated ZNT1 protein was degraded in a zinc-deficient culture containing Chelex-treated FCS, which is the same condition that facilitated ZIP4 protein accumulation in our previous studies (39,40,43,44). Therefore, zinc-responsive ZNT1 protein expression is probably physiologically significant. The regulation of ZNT1 expression on the cell surface is reciprocal to that of ZIP4. Moreover, ZIP4 is located on the apical membrane of enterocytes and takes zinc up from the intestinal lumen (45)(46)(47)(48)(49), whereas ZNT1 is located on the basolateral membrane in enterocytes and exports zinc into the bloodstream (6,10,20). We confirmed that the ZNT1 protein located on the basolateral membrane of CaCo2 cells is N-glycosylated (see Fig. 2I), which is required for fine-tuned regulation; therefore, ZNT1 expression could be controlled in a sophisticated manner in enterocytes. Reciprocal expression of ZNT1 and ZIP4 in enterocytes would contribute to zinc absorption regulation (50).
In conclusion, the results of this study dissect the molecular properties of ZNT1, and indicate that cellular and systemic zinc homeostasis is sophisticatedly maintained by the dynamic regulation of ZNT1 expression in multiple manners, although fur-

Zinc-responsive ZNT1 expression
ther studies are required to elucidate the physiological and pathological significance of ZNT1 expression regulation. Zinc entry into the cells is carried out by numerous ZIP proteins, whereas the zinc efflux pathway is mostly mediated by only ZNT1. Thus, the present study provides the molecular basis to understand homeostatic control of zinc.

Plasmid construction
Plasmids used to express N-terminally FLAG-tagged ZNT1 (both WT and mutants) were constructed by inserting each cDNA into a pA-Puro vector as described previously (26). Substitution mutants were constructed by two-step polymerase chain reaction (PCR) methods. All plasmids were linearized with appropriate restriction enzymes prior to electroporation for DT40 cells or lipofection using Lipofectamine 2000 reagent (Invitrogen) for other cells used in this study, to establish stable transfectants.

Disruption of ZNT1 genes in HAP1 and PANC1 cells
ZNT1-knockout cells were generated using the CRISPR/ Cas9 system. Guide RNA expression plasmids were generated by inserting the following oligonucleotides into the BbsI or BsaI site of a PX-330-B/B vector (53): ZNT1-F, 5Ј-CACCGGA-TCCGAGCCGAGGTAATG-3Ј; ZNT1-R, 5Ј-AAACCATTA-CCTCGGCTCGGATCC-3Ј. The constructed plasmids were cotransfected with pcDNA6/TR, which contained a blasticidin resistance gene, or pA-Neo, which contained a neomycin resistance gene, into HAP1 or PANC1 cells using Lipofectamine 2000 reagent. The cells were cultured with 20 g/ml of blasticidin (InvivoGen) or 600 g/ml of G418 sulfate (Nacalai Tesque) to generate stable clones. HAP1 or PANC1 cells deficient in the ZNT1 gene were confirmed by directly sequencing the PCR-amplified fragments using genomic DNA as a template (Table S1 and Fig. S1).

Cell surface biotinylation assay
Cell surface biotinylation assay was performed as described previously (26,43). Cells were washed with ice-cold PBS, and then EZ-Link, a sulfo-NHS-SS-biotin reagent (Pierce Protein Biology, Thermo Fisher Scientific, Rockford, IL), was added to biotinylate lysine residues exposed on the extracellular surface. Biotinylated proteins were recovered from streptavidin-coupled beads in 6ϫ SDS sample buffer and then immunoblotted.

PNGase F digestion
Digestion of N-glycosylation was performed using PNGase F (New England Biolabs, Beverly, MA). Briefly, total cell lysates were denatured with denaturing buffer (0.5% SDS, 40 mM DTT) at 37°C for 30 min and digested with PNGase F in the reaction buffer (50 mM sodium phosphate, 1% Nonidet P-40) for 2 h at 37°C. Each sample was mixed with 6ϫ SDS sample buffer and then subjected to immunoblotting. Digestion of a biotinylated membrane prepared from CaCo2 cells was performed in the presence of a protease inhibitor mixture (Nacalai Tesque) for 24 h at 37°C after denaturation.

Immunofluorescence staining
Immunostaining for the detection of ZNT1 was performed as previously described (26). Briefly, cells were fixed with 10% formaldehyde neutral buffer solution (Nacalai Tesque) and stained with anti-ZNT1 (1:1,000 dilution) followed by Alexa 488-conjugated goat anti-mouse IgG (Molecular Probes, Eugene, OR) without permeabilization. The stained cells were observed using a FSX100 fluorescent microscope (Olympus, Tokyo, Japan), and images were analyzed using Adobe Photoshop CS. Z-stack images of the immunofluorescence microscopy were also taken using FSX100.

Zinc-responsive ZNT1 expression Cytotoxicity assay against high zinc concentrations
An AlamarBlue assay was performed as previously described (26). DT40 cells were cultured at a density of 10 ϫ 10 4 cells/ml in a 96-well plate and treated with ZnSO 4 at the indicated concentrations for 2 days. PANC-1 or mtf-1 KO MEF cells were cultured at a density of 1.0 ϫ 10 4 cells/ml in a 96-well plate and treated with ZnSO 4 at the indicated concentrations for 2 days. AlamarBlue reagent was added to the medium, which was then incubated for 4 h. Absorbance was measured at 570 and 600 nm using PowerScan4.

Reverse transcription-quantitative PCR (RT-qPCR)
RT-qPCR was performed as described previously (43). Briefly, total RNA (1 g) isolated from HepG2 cells was reversetranscribed using ReverTra Ace (Toyobo, Osaka, Japan), and a real-time PCR was performed using a Thunderbird SYBR qPCR Mix (Toyobo). Samples were denatured at 95°C for 2 min and amplified for 40 reaction cycles with denaturation at 95°C for 15 s, annealing at 57°C for 15 s, and extension at 72°C for 30 s per cycle. Each cDNA sample was prepared in triplicate, and the same reaction was performed without using reverse transcriptase as a negative control. Cyclophilin A was used to normalize each sample. Primer sequences for ZNT1 and Cyclophilin A were designed as described elsewhere (55) (Table S2).

Statistical analyses
All data are presented as the mean Ϯ S.D. Statistical significance was determined using Student's t test and accepted at p Ͻ 0.05.
Author contributions-Y. N. and T. K. conceptualization; Y. N. and T. K. resources; Y. N. and T. K. data curation; Y. N. software; Y. N. formal analysis; Y. N. and T. K. funding acquisition; Y. N. and T. K. validation; Y. N. and T. K. investigation; Y. N. and T. K. visualization; Y. N. and T. K. methodology; Y. N. and T. K. writing-original draft; Y. N. and T. K. project administration; Y. N. and T. K. writing-review and editing; T. K. supervision.